EnglishVersion  
     
 
小组动态

WHU-Hi: UAV-borne hyperspectral and high spatial resolution (H2)

benchmark datasets for crop precise classification

Abstract

  WHU-Hi dataset (Wuhan UAV-borne hyperspectral image) is collected and shared by the RSIDEA research group of Wuhan University, and it could serve as a benchmark dataset for precise crop classification and hyperspectral image classification studies. The WHU-Hi dataset contains three individual UAV-borne hyperspectral datasets: WHU-Hi-LongKou, WHU-Hi-HanChuan, and WHU-Hi-HongHu. All the datasets were acquired in farming areas with various crop types in Hubei province, China, via a Headwall Nano-Hyperspec sensor mounted on a UAV platform. Compared with spaceborne and airborne hyperspectral platforms, unmanned aerial vehicle (UAV)-borne hyperspectral systems can acquire hyperspectral imagery with a high spatial resolution (which we refer to here as H2 imagery). The research was published in Remote Sensing of Environment.

1. WHU-Hi dataset

  The WHU-Hi dataset preprocessing included radiometric calibration and geometric correction, which were undertaken in the HyperSpec software provided by the instrument manufacturer. For the radiometric calibration, the raw digital number values were converted into radiance values by the laboratory calibration parameters of the sensor.

1.1 WHU-Hi-LongKou dataset

  The WHU-Hi-LongKou dataset was acquired from 13:49 to 14:37 on July 17, 2018, in Longkou Town, Hubei province, China, with an 8-mm focal length Headwall Nano-Hyperspec imaging sensor equipped on a DJI Matrice 600 Pro (DJI M600 Pro) UAV platform. During the data collection, the weather was clear and cloudless, the temperature was about 36°C, and the relative air humidity was about 65%. The study area is a simple agricultural scene, which contains six crop species: corn, cotton, sesame, broad-leaf soybean, narrow-leaf soybean, and rice. The UAV flew at an altitude of 500 m, the size of the imagery is 550 × 400 pixels, there are 270 bands from 400 to 1000 nm, and the spatial resolution of the UAV-borne hyperspectral imagery is about 0.463 m. An overview of this dataset is provided in Fig. 1 and Table. 1.

Fig. 1. The WHU-Hi-LongKou dataset. (a) Image cube. (b) Ground-truth image. (c) Typical crop photos in the study area

Table. 1. Groundtruth classes for the WHU-Hi-LongKou dataset and their respective samples number.
No. Class name Samples Legend
C1 Corn 34511
C2 Cotton 8374
C3 Sesame 3031
C4 Broad-leaf soybean 63212
C5 Narrow-leaf soybean 4151
C6 Rice 11854
C7 Water 67056
C8 Roads and houses 7124
C9 Mixed weed 5229


1.2 WHU-Hi-HanChuan dataset

  The WHU-Hi-HanChuan dataset was acquired from 17:57 to 18:46 on June 17, 2016, in Hanchuan, Hubei province, China, with an 17-mm focal length Headwall Nano-Hyperspec imaging sensor equipped on a Leica Aibot X6 UAV V1 platform. During the data collection, the weather was clear and cloudless, the temperature was about 30°C, and the relative air humidity was about 70%. The study area is a rural-urban fringe zone with buildings, water, and cultivated land, which contains seven crop species: strawberry, cowpea, soybean, sorghum, water spinach, watermelon, and greens. The UAV flew at an altitude of 250 m, the size of the imagery is 1217 × 303 pixels, there are 274 bands from 400 to 1000 nm, and the spatial resolution of the UAV-borne hyperspectral imagery is about 0.109 m. Notably, since the WHU-Hi-HanChuan dataset was acquired during the afternoon when the solar elevation angle was low, there are many shadow-covered areas in the image. An overview of this dataset is given in Fig. 2 and Table. 2.

Fig. 2. The WHU-Hi-HanChuan dataset. (a) Image cube. (b) Ground-truth image. (c) Typical crop photos in the study area

Table. 2. Groundtruth classes for the WHU-Hi-HanChuan dataset and their respective samples number.
No. Class name Samples Legend No. Class name Samples Legend
C1 Strawberry 44735
C9 Grass 9469
C2 Cowpea 22753
C10 Red roof 10516
C3 Soybean 10287
C11 Gray roof 16911
C4 Sorghum 5353
C12 Plastic 3679
C5 Water spinach 1200
C13 Bare soil 9116
C6 Watermelon 4533
C14 Road 18560
C7 Greens 5903
C15 Bright object

1136

C8 Trees 17978
C16 Water 75401


1.3 WHU-Hi-HongHu dataset

  The WHU-Hi-HongHu dataset was acquired from 16:23 to 17:37 on November 20, 2017, in Honghu City, Hubei province, China, with a 17-mm focal length Headwall Nano-Hyperspec imaging sensor equipped on a DJI Matrice 600 Pro UAV platform. During the data collection, the weather was cloudy, the temperature was about 8°C, and the relative air humidity was about 55%. The experimental area is a complex agricultural scene with many classes of crops, and different cultivars of the same crop are also planted in the region, including Chinese cabbage and cabbage, and Brassica chinensis and small Brassica chinensis. Notably, the region is planted with different cultivars of the same crop type; for example, Chinese cabbage/cabbage and brassica chinensis/small brassica chinensis. The UAV flew at an altitude of 100 m, the size of the imagery is 940 × 475 pixels, there are 270 bands from 400 to 1000 nm, and the spatial resolution of the UAV-borne hyperspectral imagery is about 0.043 m. An overview of this dataset is provided in Fig. 3 and Table. 3.

Fig. 3. The WHU-Hi-HongHu dataset. (a) Image cube. (b) Ground-truth image. (c) Typical crop photos in the study area

Table. 3. Groundtruth classes for the WHU-Hi-HongHu dataset and their respective samples number.
No. Class name Samples Legend No. Class name Samples Legend
C1 Red roof 14041
C12 Brassica chinensis 8954
C2 Road 3512
C13 Small Brassica chinensis 22507
C3 Bare soil 21821
C14 Lactuca sativa 7356
C4 Cotton 163285
C15 Celtuce 1002
C5 Cotton firewood 6218
C16 Film covered lettuce 7262
C6 Rape 44557
C17 Romaine lettuce 3010
C7 Chinese cabbage 24103
C18 Carrot

3217

C8 Pakchoi 4054
C19 White radish 8712
C9 Cabbage 10819
C20 Garlic sprout 3486
C10 Tuber mustard 12394
C21 Broad bean 1328
C11 Brassica parachinensis 11015
C22 Tree 4040


1.4 Download

  We provide three formats of WHU-Hi data. We hope you can fill in a simple questionnaire before downloading, which will appear after clicking the download link.
● Envi standard format: download
● Tiff format: download
● Matlab data format: download
● Baidu Drive (extraction code:1234):download
● Google Drive: download
● Download More:

    2.Experiment

      Table 4 shows the classification accuracy of some advanced hyperspectral classification methods on the WHU-Hi dataset.

    Table. 4. The overall accuracy (OA) of some advanced classification methods in WHU-Hi dataset
    Dataset WHU-Hi-LongKou WHU-Hi-HanChuan WHU-Hi-HongHu
    Number of training samples each class 25 50 100 25 50 100 25 50 100

    SVM[1]

    91.54

    93.23

    94.96

    61.8

    73.06

    77.61

    66.66

    67.47

    73.55

    FNEA-OO[2]

    98.37

    98.4

    98.59

    67.75

    81.02

    85.63

    85.09

    86.59

    88.83

    SVRFMC[3]

    98.18

    98.2

    98.37

    69.05

    82.02

    86.53

    84.84

    86.14

    89.86

    R-PCA CNN[4]

    92.9

    95.47

    97.3

    75.6

    84.1

    87.13

    75.63

    81.99

    85.43

    SAE-LR[5]

    84.34

    90.12

    93.78

    53.87

    60.27

    72.43

    59.21

    64.21

    78.68

    SSAN[6]

    86.9

    93.94

    94.44

    74.8

    83.63

    88.63

    72.82

    78.89

    87.34

    SSRN[7]

    91.74

    98.09

    99.02

    76.19

    83.28

    89.82

    81.17

    85.99

    91.29

    PresNet[8]

    97.78

    97.71

    98.7

    82.28

    90.19

    93.32

    86.89

    91.46

    95.32

    SSFCN-CRF[9]

    87.7

    94.28

    94.6

    73.44

    87.52

    89.75

    82.92

    90.26

    94.26

    FPGA[10]

    95.67

    97.28

    99.17

    89.28

    94.73

    97.83

    91.33

    95.76

    97.45

    CNNCRF[11]

    97.31

    97.62

    98.91

    86.94

    90.71

    93.95

    84.84

    91.06

    93.74


    Reference
     1. C.-C. Chang, and C.-J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans Intell Syst Technol, vol. 2, no. 3, pp. 27, May. 2011.
     2. M. Baatz, and A. Schäpe, “Multiresolution segmentation:an optimization approach for high quality multi-scale image segmentation,” 2000.
     3. Y. Zhong, X. Lin, and L. Zhang, “A support vector conditional random fields classifier with a Mahalanobis distance boundary constraint for high spatial resolution remote sensing imagery,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 7, no. 4, pp. 1314-1330, Apr. 2014.
     4. K. Makantasis, K. Karantzalos, A. Doulamis et al., “Deep supervised learning for hyperspectral data classification through convolutional neural networks,” Proc. IEEE IGARSS, pp. 4959-4962, Jul. 2015.
     5. Y. Chen, Z. Lin, X. Zhao et al., “Deep learning-based classification of hyperspectral data,” IEEE J Sel Top Appl Earth Observ Remote Sens, vol. 7, no. 6, pp. 2094-2107, Jun. 2014.
     6. X. Mei, E. Pan, Y. Ma et al., “Spectral-Spatial Attention Networks for Hyperspectral Image Classification,” Remote Sens, vol. 11, no. 8, pp. 963, Apr. 2019.
     7. Z. Zhong, J. Li, Z. Luo et al., “Spectral–spatial residual network for hyperspectral image classification: A 3-D deep learning framework,” IEEE Trans Geosci Remote Sens, vol. 56, no. 2, pp. 847-858, Feb. 2018.
     8. M. E. Paoletti, J. M. Haut, R. Fernandez-Beltran, J. Plaza, A. J. Plaza and F. Pla, "Deep pyramidal residual networks for spectral-spatial hyperspectral image classification", IEEE Trans. Geosci. Remote Sens., vol. 57, no. 2, pp. 740-754, Feb. 2019.
     9. Y. Xu, B. Du and L. Zhang, "Beyond the patchwise classification: Spectral–spatial fully convolutional networks for hyperspectral image classification", IEEE Trans. Big Data, Jun. 2020.
     10. Z. Zheng, Y. Zhong, A. Ma and L. Zhang, "FPGA: Fast Patch-Free Global Learning Framework for Fully End-to-End Hyperspectral Image Classification," in IEEE Trans. Geosci. Remote Sens, vol. 58, no. 8, pp. 5612-5626, Aug. 2020, doi: 10.1109/TGRS.2020.2967821.
     11. Zhong Y, Hu X, Luo C, et al. WHU-Hi: UAV-borne hyperspectral with high spatial resolution (H2) benchmark datasets and classifier for precise crop identification based on deep convolutional neural network with CRF[J]. Remote Sens. Environ, 2020, 250: 112012.


    3.Copyright

      The copyright belongs to Intelligent Data Extraction, Analysis and Applications of Remote Sensing(RSIDEA) academic research group, State Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing (LIESMARS), Wuhan University. The WHU-Hi dataset can be used for academic purposes only and need to cite the following paper, but any commercial use is prohibited. Otherwise, RSIDEA of Wuhan University reserves the right to pursue legal responsibility.

    Reference:
    [1] Y. Zhong, X. Hu, C. Luo, X. Wang, J. Zhao, and L. Zhang, “WHU-Hi: UAV-borne hyperspectral with high spatial resolution (H2) benchmark datasets and classifier for precise crop identification based on deep convolutional neural network with CRF”, Remote Sens. Environ., vol. 250, pp. 112012, 2020.
    [2] Y. Zhong, X. Wang, Y. Xu, S. Wang, T. Jia, X. Hu, J. Zhao, L. Wei, and L. Zhang, "Mini-UAV-borne hyperspectral remote sensing: From observation and processing to applications", IEEE Geosci. Remote Sens. Mag., vol. 6, no. 4, pp. 46-62, Dec. 2018.


    4.Contact

      If you have any the problem or feedback in using WHU-Hi dataset, please contact:
      Mr. Xin Hu: whu_huxin@whu.edu.cn
      Dr. Xingyu Wang: wangxinyu@whu.edu.cn
      Prof. Yanfei Zhong: zhongyanfei@whu.edu.cn

     
    版权所有@RS-IDEA | 地址:武汉市珞喻路129号 | 单位:武汉大学测绘遥感信息工程国家重点实验室 | 办公室:2教308 | zhongyanfei@whu.edu.cn